A Systematic Review of Magnesium Sulfate for Perinatal Neuroprotection: What Have We Learnt From the Past Decade?

(1)

SYSTEMATIC REVIEW published: 27 May 2020 doi: 10.3389/fneur.2020.00449

Frontiers in Neurology | www.frontiersin.org 1 May 2020 | Volume 11 | Article 449

Edited by:

Alberto Spalice, Policlinico Umberto I, Italy

Reviewed by:

Angelina Kakooza-Mwesige, Makerere University, Uganda Donna M. Ferriero, University of California, San Francisco, United States

*Correspondence:

Robert Galinsky [email protected]

Specialty section:

This article was submitted to Pediatric Neurology, a section of the journal Frontiers in Neurology

Received:11 March 2020 Accepted:28 April 2020 Published:27 May 2020

Citation:

Galinsky R, Dean JM, Lingam I, Robertson NJ, Mallard C, Bennet L and Gunn AJ (2020) A Systematic Review of Magnesium Sulfate for Perinatal Neuroprotection: What Have We Learnt From the Past Decade?

Front. Neurol. 11:449.

doi: 10.3389/fneur.2020.00449

A Systematic Review of Magnesium Sulfate for Perinatal Neuroprotection:

What Have We Learnt From the Past Decade?

Robert Galinsky^1,2*, Justin M. Dean², Ingran Lingam³, Nicola J. Robertson³, Carina Mallard⁴, Laura Bennet²and Alistair J. Gunn²

1Department of Obstetrics and Gynecology, The Ritchie Centre, Hudson Institute of Medical Research, Monash University, Melbourne, VIC, Australia,²Department of Physiology, University of Auckland, Auckland, New Zealand,³Neonatology, Institute for Women’s Health, University College London, London, United Kingdom,⁴Department of Neuroscience and Physiology, University of Gothenburg, Sahlgrenska Academy, Gothenburg, Sweden

There is an important unmet need to improve long term outcomes of encephalopathy for preterm and term infants. Meta-analyses of large controlled trials suggest that maternal treatment with magnesium sulfate (MgSO₄) is associated with a reduced risk of cerebral palsy and gross motor dysfunction after premature birth. However, to date, follow up to school age has found an apparent lack of long-term clinical benefit. Because of this inconsistency, it remains controversial whether MgSO4offers sustained neuroprotection.

We systematically reviewed preclinical and clinical studies reported from January 1 2010, to January 31 2020 to evaluate the most recent advances and knowledge gaps relating to the efficacy of MgSO4 for the treatment of perinatal brain injury. The outcomes of MgSO4 in preterm and term-equivalent animal models of perinatal encephalopathy were highly inconsistent between studies. None of the perinatal rodent studies that suggested benefit directly controlled body or brain temperature. The majority of the studies did not control for sex, study long term histological and functional outcomes or use pragmatic treatment regimens and many did not report controlling for potential study bias. Finally, most of the recent preterm or term human studies that tested the potential of MgSO₄ for perinatal neuroprotection were relatively underpowered, but nevertheless, suggest that any improvements in neurodevelopment were at best modest or absent.

On balance, these data suggest that further rigorous testing in translational preclinical models of perinatal encephalopathy is essential to ensure safety and best regimens for optimal preterm neuroprotection, and before further clinical trials of MgSO4for perinatal encephalopathy at term are undertaken.

Keywords: neuroprotection, brain injury, hypoxic-ischemic encephalopathy, perinatal encephalopathy, magnesium sulfate, cerebral palsy

(2)

Galinsky et al. Magnesium Sulfate for Perinatal Neuroprotection

INTRODUCTION

Perinatal encephalopathy is a significant public health issue.

It is associated with high mortality, serious morbidity and significant costs. Most of this cost is related to long-term neurodevelopmental disabilities, such as cerebral palsy (CP). For example, in 2003 the combined lifetime economic cost of all cases of CP in the USA was over US$ 11.5 billion (1). This cost is largely attributed to the combined loss of productive members of society and the direct burden of care on the individual, family and social institutions, rather than medical costs.

Therapeutic hypothermia for moderate to severe neonatal encephalopathy (NE) in term and near-term infants improves survival without disability (2), but nearly half of all infants still died or survived with disability. Furthermore, infants born at <35 weeks of gestation are not currently candidates for therapeutic hypothermia, and so effective therapeutic interventions to improve long-term neurodevelopmental outcomes are needed (3, 4). The etiology of the long-term disability associated with prematurity is complex, and commonly includes factors such as exposure to both antenatal and postnatal infection/inflammation where therapeutic hypothermia may not be suitable (4).

Magnesium sulfate (MgSO4) is one of the most commonly used drugs in obstetric medicine. Its first reported use was in 1925 for control of eclamptic seizures (5). An early meta-analysis of MgSO4 for preventing threatened preterm labor showed it was ineffective at preventing preterm birth, and was associated with an increased risk of infant death (6). This analysis included studies of varied gestational ages, from<32 weeks up to 37 weeks, and included women that received a 4–6 g loading dose followed by a maintenance infusion of 1.5–6 g/hour intravenously or orally. More recent meta-analyses suggest that maternal MgSO4

is associated with a small but significant reduction in the risk of CP and motor dysfunction after premature birth (number needed to treat=64), but it had no effect on overall death or disability (7,8). These meta-analyses included women in labor or at high risk of labor at≤33 weeks of gestation that received i.v. loading doses of 4–6 g of MgSO4and/or a 1–3 g/hour maintenance dose or 5 g every 4 h IM. Based on these findings, MgSO4is offered to mothers at high risk of preterm delivery in many countries (9).

Subsequently, there has been considerable interest in MgSO4

as a therapeutic intervention for preventing or mitigating the impact of perinatal brain injury (perinatal neuroprotection). In many western countries, MgSO4is now a standard treatment in pregnancies threatened by preterm labor at<30 weeks’ gestation.

Furthermore, a clinical trial is currently in progress to assess whether giving antenatal MgSO4 to women at risk of preterm labor between 30 and 34 weeks of gestation improves outcomes at 2 years of age (ACTRN12611000491965) (10). MgSO4 is one of the cheapest options for potential neuroprotection and is associated with a relatively low risk of adverse effects, and so would be highly cost effective (11). Thus, its use in the developing and developed world would be highly justified, provided long- term benefit can be demonstrated.

Currently, two out of the five randomized controlled trials of maternal MgSO4 for threatened preterm labor from the original

meta analyses have followed children up to school age and show no significant improvement in cognitive, behavioral, growth or functional outcomes, although these studies are relatively small due to incomplete follow-up (12, 13). Nevertheless, the lack of apparent long-term clinical benefit, even allowing for the smaller cohorts, raises the possibility that treating all pregnancies threatened by preterm labor may not be appropriate in the long- term. It is interesting to note that the exact mechanism/s by which MgSO4may confer neuroprotection remains unclear (14). Thus, in order to elucidate which patient population may benefit from MgSO4 we must first carefully reflect on how it interacts with evolving brain injury.

The purpose of this review was to systematically evaluate preclinical and clinical studies on MgSO4 for perinatal neuroprotection (i.e., to reduce preterm and term brain injury) over the last decade and identify the most recent advances and knowledge gaps.

ANALYSIS STRATEGY Search Method

In this review, a group of researchers focusing on perinatal neuroprotection examined preclinical (animal) and clinical (human) studies of MgSO4 for preterm and term neuroprotection in from January 1 2010, to March 1 2020 using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines for systematic review.

Studies were searched for on Pubmed and Medline (OvidSP) using the following search terms: (preterm brain injury OR perinatal encephalopathy OR neonatal encephalopathy) AND (magnesium OR MgSO4). Other sources used to identify studies included relevant original manuscripts and reviews.

Abstracts were initially identified and screened by an unbiased investigator (RG) from our study group and duplicated by another investigator (JD). Duplicate records identified from the different sources were removed (Figure 1). Full text articles were then obtained using Papers software (version 3.4.23;

Labtiva Inc., Cambridge, MA, USA) for review and independent data extraction.

The studies were grouped into those that reported improved or unaffected/negative outcomes based on histological and/or functional assessments. Studies were further subdivided by species, age (preterm vs. term equivalent), type of insult to induce brain injury, timing of treatment (before and/or after the insult), extent of temperature monitoring (brain, core, ambient, or none), main study outcomes (histological, physiological and/or behavioral), survival time after treatment, sex and study bias.

Selection Criteria

Selection of relevant studies was by group consensus. Preclinical studies were deemed to be eligible if they presented clear histological and/or functional outcomes from in vivo experiments. Histopathological outcomes were based on assessment of gray and/or white matter cell survival or tissue death. Human studies were eligible if they presented clear pathological or functional outcomes using MRI, ultrasound, CT or neurodevelopmental assessment.

(3)

FIGURE 1 |Flow chart illustrating the number of papers identified through database searching and other relevant sources, the number of full-text articles screened, assessed, and excluded, and the final number of original papers surveyed. Preclinical publications that performed ontogeny assessment of outcomes, used more than one paradigm of encephalopathy or multiple treatment timings were further sub-divided if outcomes differed according to age at assessment, experimental paradigm or treatment timing. After subdividing these publications, there was a revised total of 22 preclinical studies. For the purpose of reporting on the preclinical literature we have summarized the data based on the individual studies (n=22).

Inclusion criteria for the preclinical and human studies were preterm or term equivalentin vivostudies. We excludedin vitro studies, those that did not meet the age criteria and paradigms of traumatic brain injury. Studies that were not designed to test the effect of MgSO4 for fetal or neonatal neuroprotection as a primary outcome measure were excluded to ensure that we screened the most relevant literature and were not reporting potential false positive or false negative information.

Risk of Bias in Individual Studies

The methodological quality/risk of bias of the studies was assessed according to whether temperature was controlled, inclusion of males and females, insult and treatment were randomized and investigators were blinded to the intervention during histological/functional assessments. For human studies, additional quality criteria included whether the study was a prospective or observational and whether blinding was reported.

The results are summarized inTables 1–3.

RESULTS

We identified 195 records. After excluding reviews and records for which a full text was not available, we screened a total of 144 full text articles. One hundred and nineteen were excluded due to one or more of the following: inappropriate developmental

age, ex vivo studies, histological and/or behavioral outcomes were not examined, MgSO4 was not used for fetal or neonatal neuroprotection or assessment of MgSO4 for fetal or neonatal neuroprotection was a secondary outcome measure. Thus, a total of 25 individual publications, consisting of 16 preclinical and 9 clinical publications, were included in this analysis (Figure 1).

Preclinical publications that performed ontogeny assessment of outcomes, used more than one paradigm of encephalopathy or multiple treatment timings were further sub-divided if outcomes differed according to age at assessment, experimental paradigm or treatment timing. After subdividing these publications, there was a revised total of 22 preclinical studies. For the purpose of reporting on the preclinical literature we have summarized the data based on the individual studies (n = 22). Studies were then stratified by species (rodent, large animals, human), fetal or postnatal intervention, and outcome (neuroprotection or no neuroprotection). For clarity, rodent studies that used an antenatal intervention are referred to as “fetal rodent studies,”

whereas rodent studies that used a postnatal intervention are referred to as “postnatal rodent studies.”

Preclinical Studies of MgSO

₄

for Perinatal Neuroprotection

Most of the fetal rodent studies used acute maternal LPS injection, leading to a peak maternal systemic inflammatory

(4)

Galinskyetal.MagnesiumSulfateforPerinatalNeuroprotection TABLE 1 |Studies of MgSO4for preterm and term-equivalent neuroprotection reporting improved histological and/or functional outcomes.

References Species Insult Regime Timing Serum

levels

T^◦ Pathology Function Survival Sex Random-ization/

blinding

(15) Fetal rat, e20

Maternal LPS 270 mg/kg bolus+27 mg/kg maintenance,+270 mg/kg bolus, maternal s.c.

−2 to 2 h ? Ambient: 37^◦C during the study, data not shown

↓global expression of Casp-3, NF-KB, nNOS, IL-6 & TNF

None 4 h post

LPS

? X/?

(16) Fetal rat, e17

Maternal LPS 270 mg/kg bolus+27 mg/kg maintenance+betamethasone s.c.

+30 min to+4 h

? ? ↑% hippocampal NeuN

staining,↑mRNA expression of MAP2, MBP in♂LPS+MgSO4

vs.♂LPS+vehicle

None P 60 ♀,♂ X/?

(17) Fetal rat, e18

Maternal LPS 270 mg/kg bolus, maternal s.c. −2 to 2 h ? Ambient: 37^◦C during the study, data not shown

MRI:↓diffusivity and T2 intensities vs. control, No effect on tissue IL-1β protein expression.

None P 25 ♀ X/?

(18) Fetal rat, e16

Maternal LPS 270 mg/kg bolus+27 mg/kg maintenance,+270 mg/kg bolus, maternal s.c

−2 to 2 h ? Ambient: 37^◦C during the study, data not shown

↓brain phospho-nNOS, NF?B and CCL2 expression vs.

LPS+vehicle

None 4 h post

LPS

? X/?

(19) Fetal rat, e18

−2 to +2 h

? Ambient: 37^◦C during the study, data not shown

None ↑learning and

memory vs.

LPS+vehicle

P 90 ♂ X/?

(20) Fetal mice, e15

0 h ? ? ↓brain S100B protein

expression vs.

LPS+vehicle

None 4–6 h post

LPS

? X/?

(21) Fetal rat, e19

Bilateral UAO, 30 min

600mg/kg, maternal i.p. −20 min ? ? ↓global lipid peroxidation

and↑mitochondrial integrity score

None 60 min

post LPS

? X/X

(22) Fetal rat, e19

Bilateral UAO, 20 min

600 mg/kg, maternal i.p. −30 min ? ? ↓global lipid peroxidation

vs. vehicle+UAO

None 30 min

post LPS

? ?/?

(23) Mouse, P5 HI, 8% O2, 45 min

600 mg/kg i.p. −1 h ? Ambient: 36^◦C during

hypoxia, data not shown

↓lesion area: 55% in HI+vehicle vs. 6% in MgSO4+HI,p<0.01

Improved motor function

5 d ♀,♂ ?/?

(23) Mouse, P5 HI, 8% O2, 45 min

600 mg/kg i.p. −1 h ? Ambient: 36^◦C during

hypoxia, data not shown

Sub group analysis showed↓lesion area in the hippocampus and thalamus in♂MgSO4+HI vs.♂vehicle+HI

No

improvements in motor or cognitive function

40 d ♀,♂ ?/?

(24) Rat, P6 HI: 6% O2, 60 min

100 mg/kg i.p. −1 h 1.8

mmol/L

Ambient: 37^◦C during hypoxia, data not shown

↑MBP score,

↓microgliosis and apoptosis, no effect on oligodendrocyte survival

None 3–5 d ♂ ?/X

(25) Rat, P4 HI+10% O2

80 min

1.1 mg/g i.p. −24 h 4.1±0.2

mmol/L

Ambient: 36^◦C during hypoxia and 10 min recovery, data not shown

↓%tissue loss in hippocampal and striatal gray and white matter

None 7 d ♀♂ X/X

(Continued)

FrontiersinNeurology|www.frontiersin.org4May2020|Volume11|Article449

(5)

TABLE1|Continued ReferencesSpeciesInsultRegimeTimingSerum levelsT◦PathologyFunctionSurvivalSexRandom-ization/ blinding (25)Mouse,P5HI+10%O2 70min0.92mg/gi.p.−24h2.7±0.3 mmol/LAmbient:36◦Cduring hypoxiaand10min recovery,datanotshown

↓%hippocampaltissue lossvs.vehicle+HINone7d♀♂X/X (26)Rat,P7HI+8%O2 60min1.1mg/gi.p.−6d,−3 d,−1d, −12h,

3.3 mmol/LAmbient:36◦Cduring hypoxiaand10min recovery,datanotshown

↓globalbraininjuryscore and%tissuelossvs. vehicle+HI None7d♀♂X/X (27)Fetal sheep, 0.85g.a.

1hpartialUCO, thenneonatal resuscitation 400mg/kg,fetali.v.0h?Core:37–39◦C,datanot shown

↓corticalandsubcortical apoptosis,intracellular ROSandCa2+ accumulationvs.vehicle +UCO None3h?X/X CAO,Carotidarteryocclusion;EEG,electroencephalography;e,embryonicday;g.a.,gestationalage;HI,hypoxia-ischemia;HT,hypothermia;LPS,lipopolysaccharide;MRS,magneticresonancespectroscopy;P,postnatalday; T◦,temperature;Tx,treatment;UAO,uterinearteryocclusion;UCO,umbilicalcordocclusion.

response within∼6 h (38). All of the postnatal rodent studies used the Rice-Vannucci model of unilateral carotid artery ligation followed by a period of moderate hypoxia (39). This functional paradigm of perinatal brain injury is associated with infarction of the middle cerebral artery region ipsilateral to the ligated carotid artery, with no overt cell death in the contralateral cerebral hemisphere. In sheep, cerebral injury was caused by either partial umbilical cord occlusion or complete umbilical cord occlusion, leading to either a basal ganglion-predominant or a watershed pattern, as previously reviewed (40). Similar regional brain injury is observed in neonatal piglets following hypoxia ischemia (HI;

cerebral hypoperfusion with hypoxia) (41,42).

Half of the studies (11/22) used fetal rats or mice (e15- 20), 3/11 assessed outcomes postnatally (16, 17, 19). These gestational ages at the time of the insult broadly correspond to the neural development of human infants at<24 weeks of gestation (43, 44). Most of these studies measured markers of neuroinflammation and apoptosis, one study measured injury based on tissue S100B expression, one study used diffusion weighted and T2 magnetic resonance imaging (MRI) to broadly assess microstructural pathology, and 2 studies assessed behavior for motor or cognitive performance. Four studies quantified global expression of markers related to white matter integrity, inflammation or oxidative stress, none performed quantitative assessment of oligodendrocyte survival or myelin density. Most of the fetal rodent studies used maternal LPS to induce encephalopathy. Two fetal rat studies used bilateral uterine artery occlusion and one used mifepristone (RU486) as a model of non-inflammation associated preterm birth.

Eight out of 22 studies (36%) used postnatal rodents and induced injury using the Rice-Vannucci model. Five studied rats at P3–P6, which is broadly equivalent to the preterm human brain at 24–32 weeks of gestation, and 3 studied rats at P7 which is comparable to the preterm human at 30–32 weeks of gestation (43,44). Most of the postnatal rodent studies examined infarct area to assess neuronal loss/survival/severity of the insult, one quantified oligodendrocyte survival and 2 assessed behavior for motor and cognitive function.

Three out of 22 studies (14%) used large animals. In the large animal studies, fetal sheep were exposed to global hypoxia via umbilical cord occlusion, whereas piglets received bilateral carotid artery occlusion combined with inhalational hypoxia.

One study used fetal sheep at 0.8 (120 out of 147 days) of gestation and one used 1 day old piglets; both paradigms are comparable to the term human (45, 46). A separate study used fetal sheep at 0.7 (104 out of 147 days) of gestation, which is equivalent to human brain development at∼30 weeks of gestation (47). Histological outcomes were combined with assessment of neuronal functional and seizures in preterm fetal sheep and term piglets using electroencephalography. The term piglet study also used magnetic resonance spectroscopy to assess brain energy metabolism as one of the primary outcome measures.

Regime and Timing of Magnesium Delivery

Most of the rodent studies that used a fetal intervention (fetal rodent studies; 10/11) administered MgSO4to the pregnant dam

(6)

Galinskyetal.MagnesiumSulfateforPerinatalNeuroprotection TABLE 2 |Studies of MgSO4 for preterm and term-equivalent neuroprotection reporting no improvement in histological and/or functional outcomes.

References Species Insult Regime Timing Serum

levels

T^◦ Pathology Function Survival Sex Random-ization/

blinding

(16) Fetal rat e17

Maternal LPS 270 mg/kg bolus+27 mg/kg maintenance+BM s.c.

+30 min – 4 h

? ? None No effect on

motor function at P13 despite improvements at P5 and P9 in

♂LPS+MgSO4

vs.

♂LPS+vehicle

P 5, 9, 13 ♀♂ X/?

Maternal Mifepristone (RU486)

270 mg/kg bolus+27 mg/kg maintenance,+270 mg/kg bolus, maternal s.c

0 h ? ? No change in S100B

protein expression

None 4–6 h post

RU486

? X/?

0 h ? ? No effect on

inflammatory, oligodendrocyte and astrocyte gene expression.↑Caspase-1 and IL-1βmRNA expression in LPS+MgSO4vs.

vehicle+MgSO4

None 6 h post

LPS

? X/?

(26) Rat, P7 HI+8% O2

60 min

1.1 mg/g i.p. −3 h,

−30 min 3.3 mmol/L

No differences between groups for injury score and % tissue loss vs.

vehicle+HI

None 7 d ♀♂ X/?

(26) Rat, P7 HI+8% O2

60 min

1.1 mg/g i.p. 1 h 3.3

mmol/L

No differences between groups for injury score and % tissue loss vs.

vehicle+HI

None 7 d ♀♂ X/X

(29) Fetal

sheep 0.7 gestation

25 min UCO 160 mg loading, 48 mg/h/24 then 24 mg/h/24 h

−24–

+24

1.9±0.1 mmol/L

Brain: 39-39.6^◦C, continuous monitoring

↑oligodendrocyte loss.

No effect on neuronal survival

↓seizure number and burden

3 d ♂♀ X/X

(30) P1 piglets Bilateral CAO+FiO2 6%

180 mg/kg loading then 8 mg/kg/h 48 h+HT vs.

vehicle+HT

1 h 1.5

mmol/L

Core: normothermia:

38–39 vs. HT:

33.5–34^◦C, continuous monitoring

No regional improvements in cell death or oligodendrocyte survival.

↓in total cell death and↑ oligodendrocyte survival in MgSO4+HT vs. vehicle+HT

No overall improvement in recovery of EEG or MRS markers of outcome Secondary analysis suggested↑ recovery of aEEG when mild cases were excluded

2 d ♂ X/X

CAO, Carotid artery occlusion; EEG, electroencephalography; e, embryonic day; g.a, gestational age; HI, hypoxia ischemia; HT, hypothermia; LPS, lipopolysaccharide; MRS, magnetic resonance spectroscopy; P, postnatal day;

T^◦, temperature; Tx, treatment; UAO, uterine artery occlusion; UCO, umbilical cord occlusion.

(7)

Galinskyetal.MagnesiumSulfateforPerinatalNeuroprotection TABLE 3 |Human studies of MgSO4 for preterm and term/late preterm neuroprotection.

References Indication Age at birth Age at study Timing Pathology Function Sex Study type

(31) Preterm

neuroTx:

chorioamnionitis

≤30 weeks, n=228

2 y Antenatal No difference in rates of IVH

between CA+MgSO4 vs.

CA+placebo

No difference in rates of CP, mental or physical disability

X Prospective, double blinded multicenter RCT

(32) Preterm

neuroTx:

chorioamnionitis

≥24 weeks n=396

2 y Antenatal No difference in rates of IVH

or PVL between CA+MgSO4 vs.

CA+placebo

No difference in rate of stillbirth, death, moderate-severe CP or neurodevelopmental delay between groups

(33) Preterm

neuroTx

Mean±SD: 28.3

±2.2 weeks, n=73

Mean±SD:

32.4±2 weeks

Antenatal MRI:↓cerebellar hemorrhage in MgsO4 (n=13/49) vs. control (n=14/24), no effect on white matter injury or IVH

None X Prospective, blinded single

center study

(34) Preterm

neuroTx

Median±SD: 27

±2, weeks, n=64

24, 48, 72 h Antenatal Cranial ultrasound:↓P/IVH in MgSO4 (n=4/36) vs.

vehicle (n=9/28) at 72 h

NIRS:↓cerebral O2

consumption vs. vehicle at 24 h

X Observational single center study

(13) Preterm

neuroTx

Mean±SD: 27.3

±2.2, weeks n=867

7–8 years old Antenatal None No difference in cognitive,

behavioral, growth or functional outcomes

(12) Preterm

neuroTx

27–32 weeks n=501

7–14 years old Antenatal None No significant improvement

in motor dysfunction, behavior or cognition

(35) Preterm

neuroTx

24–32 weeks, n=475

Hospital discharge Antenatal No difference in rates of IVH or PVL. Secondary outcomes showed higher incidence of ROP. Subgroup analysis showed higher neonatal mortality rate with increasing magnesium levels

MgSO4group took longer to reach full feeds and had greater length of hospital stay vs. placebo

X Retrospective study

(36) Term HIE ≥35 weeksn=60 Not stated,

assessed at hospital discharge

≤6 h after birth

No difference in HIE severity, intracranial hemorrhage or death between MgSO4+HT (n=29) vs. vehicle+HT (n=31)

No difference in seizures χ Prospective, double blinded multicenter RCT

(37) Term HIE 38–39 weeks,

n=32

Follow up to 6 months

≤30 min, 24 and 48 h after birth

Brian CT: No difference in severity of HIE in MgSO4

(n=16) vs. control (n=16)

No difference in incidence of death, seizures,

discontinuous EEG or neurodevelopmental assessments at discharge and 6 months

χ Prospective single center RCT

CA, chorioamnionitis; CP, cerebral palsy; CT, computed tomography; EEG, electroencephalography; HIE, hypoxic ischemic encephalopathy; P/IVH, peri/intraventricular hemorrhage; NIRS, near infrared spectroscopy; PVL, periventricular leukomalacia; RCT, randomized controlled trial; Tx, treatment.

(8)

subcutaneously using a 270 mg/kg bolus followed by a 27 mg/kg maintenance infusion every 20 min for 4 h. An additional 270 mg/kg MgSO4 bolus was given at the end of the 4 h period in all but 1 study, in which MgSO4 and betamethasone were given 30 min after maternal lipopolysaccharide (LPS) injection.

However, in that study the effect of MgSO4alone was not tested.

In postnatal rodents, MgSO4 was administered s.c. or i.p. as a single dose that ranged from 100 to 1,000 mg/kg. In the large animal studies the dose ranged from 160 to 400 mg/kg and was delivered intravenously. Two out of 3 large animal studies (1 preterm fetal sheep and 1 term piglet paper) used an additional maintenance infusion over 48 h (range: 16–48 mg/kg).

The majority of the studies (14/22; 64%) administered MgSO4

before the insult (−24 h to−30 min), 12/14 reported that MgSO4

was associated with neuroprotection. Six out of 22 studies (27%) started treatment immediately after the insult (≤30 min), 3/6 reported MgSO4 was associated with neuroprotection. Two out of 22 studies (9%) delayed treatment to 1 h post insult, neither of these studies reported significant improvements in histological and/or functional outcomes.

Neurological Outcome and Survival Time After Injury

Fifteen out of the 22 perinatal studies (68%) reported improved neural outcomes with MgSO4 treatment. One of these studies reported improved outcomes when MgSO4 was combined with betamethasone in the setting of maternal inflammation, although the effects of MgSO4 alone were not assessed. Intriguingly, in sham controls combination therapy was associated with loss of NeuN immunoreactivity, indicating reduced neuronal survival (16). Eleven out of 15 studies (73%) used relatively short survival times of<1 d (n=6) or 1–7 d (n=5). One out of 15 studies (7%) examined outcomes after 2–4 weeks. Three out of 15 studies (20%) used survival times of 5–13 weeks. Two out of 3 studies, that used survival times of 35 and 64 days, showed histological injury and improvements with treatment were limited to males (there were no long-term effects of the insult or treatment in females) (16,23), however there were no improvements in motor or cognitive function in male offspring (23). One study, which was limited to male offspring, reported improvements in learning and memory after 93 days but did not examine histology (19).

Seven out of 22 studies (32%) reported no neuroprotection or deleterious effects associated with MgSO4 treatment. One study in fetal rodents reported no effect of MgSO4on expression of pyroptotic and pro-inflammatory genes after maternal LPS exposure, or on pro-oligodendrocyte and glial markers. However, MgSO4 treatment plus LPS was associated with increased expression of pyroptotic and pro-inflammatory gene expression compared to sham controls (28). One study assessed combination therapy with MgSO4and therapeutic hypothermia in term piglets exposed to HI and showed reduced total numbers of TUNEL positive cells and increased total numbers of oligodendrocytes compared to therapeutic hypothermia alone. However, MgSO4

was not associated with any improvements in regional cell survival or in magnetic resonance spectroscopy measures of outcome at 48 h compared with hypothermia. One study in

preterm fetal sheep exposed to severe asphyxia reported reduced seizures but no improvement in regional neuronal survival and reduced survival of oligodendrocytes in MgSO4 treated fetuses compared to vehicle+asphyxia. One study in rodents exposed to maternal LPS reported improvements in motor function at P5 and P9, however at P13 there was no effect of treatment (16). Six out of 7 studies used relatively short survival times of<1 d (n=2) or 1–7 d (n=4). One study examined outcomes after 2 weeks.

Temperature Monitoring

In the fetal rodent studies that reported improved outcomes with MgSO4, 4/8 studies controlled ambient temperature and 4 did not report temperature monitoring in their experimental protocol.

All 6 postnatal rodent studies that reported improved outcomes with MgSO4 monitored ambient temperature. However, 3/6 monitored ambient temperature during 10 min of recovery and 3/6 monitored temperature during HI but not the recovery period. One study in preterm neonatal lambs that showed improved outcomes with MgSO4 reported control of core temperature during the study period but data were not shown.

Furthermore, the core temperature range reported in this study was 37–39^◦C. Critically, the lower end of this range is∼2–3^◦C lower than the average core temperature of a neonatal lamb.

Three out of 7 studies that showed no improvement or deleterious effects with MgSO4 did not report temperature monitoring; all used rodents exposed to antenatal insults and treatment. Two studies, both in the P7 rat, reported ambient temperature control during the first 10 min of recovery. All of the large animal studies that showed no improvement/some modest improvements/deleterious effects with MgSO4 (2/7) reported continuous core or brain temperature data throughout the study period.

Sex

Six out of 15 studies (40%) that reported improved outcomes used both sexes. One of these studies in postnatal rodents exposed to HI reported histological benefits were limited to MgSO4-treated males but no improvements were seen in motor and cognitive deficits. Two studies only used males, and one study only used females. Six studies did not report the sex of the subjects.

Four out of 7 studies that reported no improvement in outcomes used both sexes. One study in neonatal piglets examined males. Two studies did not report the sex of the subjects.

Measurement of Magnesium in the Circulation and Brain

Seven out of 22 studies reported circulating magnesium concentrations in their subjects. Levels in small and large animal studies ranged from 1.5 to 4.1 mmol/L (100–350% increase from baseline). One study in piglets reported a 16 % increase in CSF magnesium levels during treatment.

Study Bias

Five out of 15 studies (33%) that showed improved outcomes reported subjects were randomized to the insult and treatment,

(9)

and blinding of assessors. Six out of 15 studies (40%) reported randomization but did not report blinding of assessors. One out of 15 studies (7%) reported blinding but not randomization of subjects to the insult or treatment. Three out of 15 studies (20%) did not report randomization or blinding as part of their experimental protocol.

Four out of 7 studies (57%) that showed no improvement/some modest improvements/deleterious effects reported randomization of subjects to the insult and treatment, and blinding of assessors during analyses. Three out of 7 (43%) studies reported randomization of subjects to the treatment and insult but did not report blinding of assessors in their protocol.

Human Publications

We identified 9 human publications. Two papers were follow- up analyses of large randomized controlled trials investigating school aged outcomes in preterm infants treated with antenatal MgSO4 (n = 503–669) (12, 13). Two papers were secondary analyses of a randomized control trial of MgSO4 for fetal neuroprotection (n= 228–396) (31,32). Five papers reported short term outcomes in preterm or term patients; 4 papers assessed a relatively small number of patients (totaln=32–71) (33, 34, 36, 37), while 1 assessed short term outcomes from a larger population (n=475) (35).

Seven papers examined preterm infants that received antenatal MgSO4for fetal neuroprotection. Two papers reported no significant improvement in cognitive, motor, behavioral, growth or functional outcomes in school age children (12,13).

Two were sub-group analyses focusing on infants exposed to clinical chorioamnionitis from a large randomized controlled trial (48); both showed MgSO4 was not associated with improved neurodevelopment at 2 years of age or reduced rates of intraventricular hemorrhage (IVH) or periventricular leukomalacia (PVL) (31, 32). Two papers reported a reduction in cerebral or cerebellar hemorrhage in the MgSO4 group vs.

placebo, however rates of hemorrhage were low and the total number of patients included was relatively small (n = 64–

71) (33, 34). One publication in 475 preterm infants reported no effect of MgSO4 on rates of IVH and PVL at hospital discharge. Secondary analyses showed higher rates of retinopathy of prematurity, longer time to reach full feeds and a higher length of hospital stay in the MgSO4 group vs. placebo (35).

Two papers examined term/near term infants; one used postnatal MgSO4as an adjunct to hypothermia (36), the other used MgSO4

alone (37). Treatment was administered within the first 30 min or 6 h after birth. There were no differences in pathological or functional outcomes between groups in both of the term human studies which were assessed at hospital discharge and 6 months of age. Neurological outcomes from the human studies were not consistent across the studies and so were not able to be compared.

Seven out of 9 papers (all preterm) reported equal ratios of males and females in their demographic data (12,13,31–35). Two studies, both in term infants, did not report patient sex (36,37).

Seven out of 9 papers were prospective (12,13,31–33,36,37), one study in preterm infants was an observational analysis (34) and 1 was a retrospective analysis (35). Two out of 9 studies

(1 preterm and 1 term) did not report blinding in their study protocol (34,37).

DISCUSSION

This systematic review shows that the effect of MgSO4

for neuroprotection in preterm and term-equivalent models of perinatal encephalopathy was highly inconsistent between studies in the last decade. Of particular concern, none of the perinatal rodent studies that suggested benefit directly controlled body or brain temperature or delay therapeutic intervention beyond the first hour after the experimental insult, whereas studies that did report control of core or brain temperature showed either limited or no improvement in outcomes. The majority of the studies did not control for sex or study long term histological and functional outcomes and many did not report appropriate measures to control for potential study bias. Moreover, although many of the recent preterm or term equivalent human studies that tested the potential of MgSO4

for perinatal neuroprotection were relatively small and likely to be underpowered, none report significant improvements in neurodevelopment, and follow up studies of the original multicenter randomized controlled trials of maternal MgSO4

for neuroprotection suggest that MgSO4 is not associated with improved outcomes at school age.

Iatrogenic Hypothermia

As previously reviewed (14), magnesium consistently promotes peripheral vasodilation in rodents, large animals and humans (49–51). Thus, a magnesium-induced increase in cutaneous perfusion is likely to increase heat loss through radiation (50);

this may be considerable in small animals with a relatively large surface area and low thermal stability (52). Supporting this concept, studies in adult rodents showed that neuroprotective effects of MgSO4 were associated with mild hypothermia and that neuroprotective effects were abolished if normothermia was maintained. In the preclinical studies surveyed here, over half (53%) of those that reported neuroprotection with MgSO4 used maternal LPS exposure to model antenatal infection/inflammation. Maternal LPS exposure commonly leads to maternal pyrexia (53), however, none of these studies reported maternal body temperature. Critically, intrapartum fever is associated with adverse neonatal outcomes (54–56) and increased risk of unexplained cerebral palsy or neonatal encephalopathy (57). Furthermore, it is well-established that mild hyperthermia of just 1 to 2^◦C can worsen neural injury (58,59), possibly via greater release of oxygen free radicals and excitatory neurotransmitters, enhanced toxicity of glutamate on neurons, greater dysfunction of the blood brain barrier and proteolysis (40, 60). Half of the studies that used maternal LPS to study MgSO4 for fetal neuroprotection did not report temperature monitoring in their study protocol, while the remaining studies reported ambient temperature monitoring during the study;

none presented temperature data in their results.

Most of the remaining studies used postnatal rodents (P5–P7) exposed to HI to evaluate MgSO4for neonatal neuroprotection.

All reported maintenance of ambient temperature during